ARCHIVES

Colorectal cancer is one of the most serious health concerns worldwide, and early diagnosis is important for improving patient survival and treatment effectiveness. This paper presents a portable and economical colorectal cancer detection system developed using deep learning and edge computing technologies. Colonoscopy images are analyzed using the lightweight MobileNetV2 neural network to identify abnormal tissues and polyp regions efficiently.The trained model is converted into TensorFlow Lite format and implemented on Raspberry Pi 5 for real-time prediction in healthcare environments with limited computing resources. Along with image-based analysis, the system monitors patient vital signs such as heart rate, oxygen saturation (SpO₂), and body temperature using MAX-series biomedical sensors. The proposed system was evaluated using publicly available colonoscopy datasets including Kvasir and HyperKvasir. Experimental results demonstrated strong classification performance with an AUC score of 0.9512, recall of 93.0%, precision of 80.87%, and MCC of 0.7181.